INTRODUCTION

In the oil and gas industry, subsurface imaging with the reflection seismic method has been performed for many decades and has been accepted as a successful approach. The procedure of subsurface imaging from raw seismic data consists of many complicated sub-processes and is often organized differently according to its various purposes. In many cases, however, human interpretation of event discontinuity that indicates faults, unconformities, and buried channels is required during processing, or at the end of the procedure. Such interpretations are usually accomplished by an experienced interpreter who has indepth knowledge of geology. Even for an experienced expert, the interpretation task is challenging if the image is a 3D volume.

In order to help with such tasks, a process called ``coherency evaluation'' has been developed Bahorich and Farmer (1995); Marfurt et al. (1998). The coherency evaluation calculates continuity of seismic events from a 3D seismic image cube and produces a so-called ``coherency cube'' that shows the distribution of the event continuity in a 3D volume. The output of coherency evaluation processing helps the interpreters to locate geologically meaningful discontuinities more easily. However, interpreting and mapping 3D discontinuity surfaces by hand is still a challenging job because of the difficulty in visualizing a 3D image and the complexity of surfaces in a 3D sense.

As a supplementary tool for mapping discontinuities, I introduce an automatic discontinuity extraction method. The proposed method starts with a coherency evaluation for a 3D seismic image and produces a discontinuity map that locates the event's discontinuities in the form of arbitarily shaped 3D surfaces. The output of the proposed method could be a good starting point for an interpreter to narrow down the various discontinuities into more meaningful geological features such as faults, unconformities, and buried channels.

In the following section I will shortly review the coherency evaluation method to clarify the meaning of a discontinuity in a seismic image. The subsequent two sections are devoted to explaining the following procedures. In each step, the algorithm is explained and also demonstrated with a real seismic image. The testing image used is from the Boonsville natural gas field located in north-central Texas and was acquired by the DOE and the Gas Research Institute as part of a secondary gas recovery technique development. According to Hardage Hardage (1996), the Boonsville seismic image is a full-fold time migrated section with $110m \times 110m$ in bin size and covers 5.5 km² area. The data consists of 97 lines along the crossline and each line has 113 traces. Figure 1 shows three plane sections of the image volume which are orthogonal to each other.

 

seis
Figure 1 Three selected sections of Boonsville 3D seismic image (time slice at t=250ms, the 37th crossline section, and the 40th inline section.)